Intro to OCO
ver 2.0
Updated optimization chapter including the Polyak stepsize rule and analysis
Learning in changing environments: adaptive regret
Boosting from OCO
Online boosting and its relationship to the Frank Wolfe method
Multi-objective games and Blackwell Approachability
New exercises and printed solutions (for instructors only)
New chapters include:
Chapter 10 on boosting for supervised learning and its relationship to online convex optimization. Derivation of boosting algorithms from regret minimization.
Chapter 11 on the notion of adaptive regret and algorithms for online convex optimization with near-optimal adaptive regret bounds.
Chapter 12 on online boosting and its relationship to the Frank-Wolfe method.
Chapter 13 on Blackwell approachability and its strong connection to online convex optimization.